perf_counter: Rename various fields
authorPeter Zijlstra <a.p.zijlstra@chello.nl>
Tue, 2 Jun 2009 13:13:03 +0000 (15:13 +0200)
committerIngo Molnar <mingo@elte.hu>
Tue, 2 Jun 2009 19:45:30 +0000 (21:45 +0200)
A few renames:

  s/irq_period/sample_period/
  s/irq_freq/sample_freq/
  s/PERF_RECORD_/PERF_SAMPLE_/
  s/record_type/sample_type/

And change both the new sample_type and read_format to u64.

Reported-by: Stephane Eranian <eranian@googlemail.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: John Kacur <jkacur@redhat.com>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/powerpc/kernel/perf_counter.c
arch/x86/kernel/cpu/perf_counter.c
include/linux/perf_counter.h
kernel/perf_counter.c

index f96d55f55bd604857a6cf019e5aceeb1c9494911..c9633321e7a58277ef88277a7097f2a2d525520e 100644 (file)
@@ -535,7 +535,7 @@ void hw_perf_enable(void)
                        continue;
                }
                val = 0;
-               if (counter->hw.irq_period) {
+               if (counter->hw.sample_period) {
                        left = atomic64_read(&counter->hw.period_left);
                        if (left < 0x80000000L)
                                val = 0x80000000L - left;
@@ -749,12 +749,12 @@ static void power_pmu_unthrottle(struct perf_counter *counter)
        s64 val, left;
        unsigned long flags;
 
-       if (!counter->hw.idx || !counter->hw.irq_period)
+       if (!counter->hw.idx || !counter->hw.sample_period)
                return;
        local_irq_save(flags);
        perf_disable();
        power_pmu_read(counter);
-       left = counter->hw.irq_period;
+       left = counter->hw.sample_period;
        val = 0;
        if (left < 0x80000000L)
                val = 0x80000000L - left;
@@ -789,7 +789,7 @@ static int can_go_on_limited_pmc(struct perf_counter *counter, u64 ev,
        if (counter->hw_event.exclude_user
            || counter->hw_event.exclude_kernel
            || counter->hw_event.exclude_hv
-           || counter->hw_event.irq_period)
+           || counter->hw_event.sample_period)
                return 0;
 
        if (ppmu->limited_pmc_event(ev))
@@ -925,7 +925,7 @@ const struct pmu *hw_perf_counter_init(struct perf_counter *counter)
 
        counter->hw.config = events[n];
        counter->hw.counter_base = cflags[n];
-       atomic64_set(&counter->hw.period_left, counter->hw.irq_period);
+       atomic64_set(&counter->hw.period_left, counter->hw.sample_period);
 
        /*
         * See if we need to reserve the PMU.
@@ -958,7 +958,7 @@ const struct pmu *hw_perf_counter_init(struct perf_counter *counter)
 static void record_and_restart(struct perf_counter *counter, long val,
                               struct pt_regs *regs, int nmi)
 {
-       u64 period = counter->hw.irq_period;
+       u64 period = counter->hw.sample_period;
        s64 prev, delta, left;
        int record = 0;
        u64 addr, mmcra, sdsync;
index 316b0c995f3878b4a8efc135472a89baad3948b0..ec06aa5e9282ec8d6a34603560017ba139786777 100644 (file)
@@ -290,11 +290,11 @@ static int __hw_perf_counter_init(struct perf_counter *counter)
        hwc->nmi        = 1;
        hw_event->nmi   = 1;
 
-       if (!hwc->irq_period)
-               hwc->irq_period = x86_pmu.max_period;
+       if (!hwc->sample_period)
+               hwc->sample_period = x86_pmu.max_period;
 
        atomic64_set(&hwc->period_left,
-                       min(x86_pmu.max_period, hwc->irq_period));
+                       min(x86_pmu.max_period, hwc->sample_period));
 
        /*
         * Raw event type provide the config in the event structure
@@ -462,7 +462,7 @@ x86_perf_counter_set_period(struct perf_counter *counter,
                             struct hw_perf_counter *hwc, int idx)
 {
        s64 left = atomic64_read(&hwc->period_left);
-       s64 period = min(x86_pmu.max_period, hwc->irq_period);
+       s64 period = min(x86_pmu.max_period, hwc->sample_period);
        int err;
 
        /*
index 4845a214b9e7a6e5eaef6dc9e347ac3798545dad..1fcd3cc93855a8e5718a2bc24381173a1a843909 100644 (file)
@@ -94,18 +94,18 @@ enum sw_event_ids {
 #define PERF_COUNTER_EVENT_MASK                __PERF_COUNTER_MASK(EVENT)
 
 /*
- * Bits that can be set in hw_event.record_type to request information
+ * Bits that can be set in hw_event.sample_type to request information
  * in the overflow packets.
  */
-enum perf_counter_record_format {
-       PERF_RECORD_IP                  = 1U << 0,
-       PERF_RECORD_TID                 = 1U << 1,
-       PERF_RECORD_TIME                = 1U << 2,
-       PERF_RECORD_ADDR                = 1U << 3,
-       PERF_RECORD_GROUP               = 1U << 4,
-       PERF_RECORD_CALLCHAIN           = 1U << 5,
-       PERF_RECORD_CONFIG              = 1U << 6,
-       PERF_RECORD_CPU                 = 1U << 7,
+enum perf_counter_sample_format {
+       PERF_SAMPLE_IP                  = 1U << 0,
+       PERF_SAMPLE_TID                 = 1U << 1,
+       PERF_SAMPLE_TIME                = 1U << 2,
+       PERF_SAMPLE_ADDR                = 1U << 3,
+       PERF_SAMPLE_GROUP               = 1U << 4,
+       PERF_SAMPLE_CALLCHAIN           = 1U << 5,
+       PERF_SAMPLE_CONFIG              = 1U << 6,
+       PERF_SAMPLE_CPU                 = 1U << 7,
 };
 
 /*
@@ -132,12 +132,12 @@ struct perf_counter_hw_event {
        __u64                   config;
 
        union {
-               __u64           irq_period;
-               __u64           irq_freq;
+               __u64           sample_period;
+               __u64           sample_freq;
        };
 
-       __u32                   record_type;
-       __u32                   read_format;
+       __u64                   sample_type;
+       __u64                   read_format;
 
        __u64                   disabled       :  1, /* off by default        */
                                nmi            :  1, /* NMI sampling          */
@@ -262,7 +262,7 @@ enum perf_event_type {
         * struct {
         *      struct perf_event_header        header;
         *      u64                             time;
-        *      u64                             irq_period;
+        *      u64                             sample_period;
         * };
         */
        PERF_EVENT_PERIOD               = 4,
@@ -363,7 +363,7 @@ struct hw_perf_counter {
                };
        };
        atomic64_t                      prev_count;
-       u64                             irq_period;
+       u64                             sample_period;
        atomic64_t                      period_left;
        u64                             interrupts;
 #endif
index 978ecfcc7aafd60d68ee7fcccda6cf67f0d4283b..5ecd9981c03579f1d7e283e391b7e620fa002676 100644 (file)
@@ -1186,7 +1186,7 @@ static void perf_log_period(struct perf_counter *counter, u64 period);
 static void perf_adjust_freq(struct perf_counter_context *ctx)
 {
        struct perf_counter *counter;
-       u64 interrupts, irq_period;
+       u64 interrupts, sample_period;
        u64 events, period;
        s64 delta;
 
@@ -1204,23 +1204,23 @@ static void perf_adjust_freq(struct perf_counter_context *ctx)
                        interrupts = 2*sysctl_perf_counter_limit/HZ;
                }
 
-               if (!counter->hw_event.freq || !counter->hw_event.irq_freq)
+               if (!counter->hw_event.freq || !counter->hw_event.sample_freq)
                        continue;
 
-               events = HZ * interrupts * counter->hw.irq_period;
-               period = div64_u64(events, counter->hw_event.irq_freq);
+               events = HZ * interrupts * counter->hw.sample_period;
+               period = div64_u64(events, counter->hw_event.sample_freq);
 
-               delta = (s64)(1 + period - counter->hw.irq_period);
+               delta = (s64)(1 + period - counter->hw.sample_period);
                delta >>= 1;
 
-               irq_period = counter->hw.irq_period + delta;
+               sample_period = counter->hw.sample_period + delta;
 
-               if (!irq_period)
-                       irq_period = 1;
+               if (!sample_period)
+                       sample_period = 1;
 
-               perf_log_period(counter, irq_period);
+               perf_log_period(counter, sample_period);
 
-               counter->hw.irq_period = irq_period;
+               counter->hw.sample_period = sample_period;
        }
        spin_unlock(&ctx->lock);
 }
@@ -2297,7 +2297,7 @@ static void perf_counter_output(struct perf_counter *counter,
                                int nmi, struct pt_regs *regs, u64 addr)
 {
        int ret;
-       u64 record_type = counter->hw_event.record_type;
+       u64 sample_type = counter->hw_event.sample_type;
        struct perf_output_handle handle;
        struct perf_event_header header;
        u64 ip;
@@ -2321,61 +2321,61 @@ static void perf_counter_output(struct perf_counter *counter,
        header.misc = PERF_EVENT_MISC_OVERFLOW;
        header.misc |= perf_misc_flags(regs);
 
-       if (record_type & PERF_RECORD_IP) {
+       if (sample_type & PERF_SAMPLE_IP) {
                ip = perf_instruction_pointer(regs);
-               header.type |= PERF_RECORD_IP;
+               header.type |= PERF_SAMPLE_IP;
                header.size += sizeof(ip);
        }
 
-       if (record_type & PERF_RECORD_TID) {
+       if (sample_type & PERF_SAMPLE_TID) {
                /* namespace issues */
                tid_entry.pid = perf_counter_pid(counter, current);
                tid_entry.tid = perf_counter_tid(counter, current);
 
-               header.type |= PERF_RECORD_TID;
+               header.type |= PERF_SAMPLE_TID;
                header.size += sizeof(tid_entry);
        }
 
-       if (record_type & PERF_RECORD_TIME) {
+       if (sample_type & PERF_SAMPLE_TIME) {
                /*
                 * Maybe do better on x86 and provide cpu_clock_nmi()
                 */
                time = sched_clock();
 
-               header.type |= PERF_RECORD_TIME;
+               header.type |= PERF_SAMPLE_TIME;
                header.size += sizeof(u64);
        }
 
-       if (record_type & PERF_RECORD_ADDR) {
-               header.type |= PERF_RECORD_ADDR;
+       if (sample_type & PERF_SAMPLE_ADDR) {
+               header.type |= PERF_SAMPLE_ADDR;
                header.size += sizeof(u64);
        }
 
-       if (record_type & PERF_RECORD_CONFIG) {
-               header.type |= PERF_RECORD_CONFIG;
+       if (sample_type & PERF_SAMPLE_CONFIG) {
+               header.type |= PERF_SAMPLE_CONFIG;
                header.size += sizeof(u64);
        }
 
-       if (record_type & PERF_RECORD_CPU) {
-               header.type |= PERF_RECORD_CPU;
+       if (sample_type & PERF_SAMPLE_CPU) {
+               header.type |= PERF_SAMPLE_CPU;
                header.size += sizeof(cpu_entry);
 
                cpu_entry.cpu = raw_smp_processor_id();
        }
 
-       if (record_type & PERF_RECORD_GROUP) {
-               header.type |= PERF_RECORD_GROUP;
+       if (sample_type & PERF_SAMPLE_GROUP) {
+               header.type |= PERF_SAMPLE_GROUP;
                header.size += sizeof(u64) +
                        counter->nr_siblings * sizeof(group_entry);
        }
 
-       if (record_type & PERF_RECORD_CALLCHAIN) {
+       if (sample_type & PERF_SAMPLE_CALLCHAIN) {
                callchain = perf_callchain(regs);
 
                if (callchain) {
                        callchain_size = (1 + callchain->nr) * sizeof(u64);
 
-                       header.type |= PERF_RECORD_CALLCHAIN;
+                       header.type |= PERF_SAMPLE_CALLCHAIN;
                        header.size += callchain_size;
                }
        }
@@ -2386,28 +2386,28 @@ static void perf_counter_output(struct perf_counter *counter,
 
        perf_output_put(&handle, header);
 
-       if (record_type & PERF_RECORD_IP)
+       if (sample_type & PERF_SAMPLE_IP)
                perf_output_put(&handle, ip);
 
-       if (record_type & PERF_RECORD_TID)
+       if (sample_type & PERF_SAMPLE_TID)
                perf_output_put(&handle, tid_entry);
 
-       if (record_type & PERF_RECORD_TIME)
+       if (sample_type & PERF_SAMPLE_TIME)
                perf_output_put(&handle, time);
 
-       if (record_type & PERF_RECORD_ADDR)
+       if (sample_type & PERF_SAMPLE_ADDR)
                perf_output_put(&handle, addr);
 
-       if (record_type & PERF_RECORD_CONFIG)
+       if (sample_type & PERF_SAMPLE_CONFIG)
                perf_output_put(&handle, counter->hw_event.config);
 
-       if (record_type & PERF_RECORD_CPU)
+       if (sample_type & PERF_SAMPLE_CPU)
                perf_output_put(&handle, cpu_entry);
 
        /*
-        * XXX PERF_RECORD_GROUP vs inherited counters seems difficult.
+        * XXX PERF_SAMPLE_GROUP vs inherited counters seems difficult.
         */
-       if (record_type & PERF_RECORD_GROUP) {
+       if (sample_type & PERF_SAMPLE_GROUP) {
                struct perf_counter *leader, *sub;
                u64 nr = counter->nr_siblings;
 
@@ -2702,7 +2702,7 @@ void perf_counter_munmap(unsigned long addr, unsigned long len,
 }
 
 /*
- * Log irq_period changes so that analyzing tools can re-normalize the
+ * Log sample_period changes so that analyzing tools can re-normalize the
  * event flow.
  */
 
@@ -2725,7 +2725,7 @@ static void perf_log_period(struct perf_counter *counter, u64 period)
                .period = period,
        };
 
-       if (counter->hw.irq_period == period)
+       if (counter->hw.sample_period == period)
                return;
 
        ret = perf_output_begin(&handle, counter, sizeof(freq_event), 0, 0);
@@ -2834,7 +2834,7 @@ static void perf_swcounter_set_period(struct perf_counter *counter)
 {
        struct hw_perf_counter *hwc = &counter->hw;
        s64 left = atomic64_read(&hwc->period_left);
-       s64 period = hwc->irq_period;
+       s64 period = hwc->sample_period;
 
        if (unlikely(left <= -period)) {
                left = period;
@@ -2874,7 +2874,7 @@ static enum hrtimer_restart perf_swcounter_hrtimer(struct hrtimer *hrtimer)
                        ret = HRTIMER_NORESTART;
        }
 
-       period = max_t(u64, 10000, counter->hw.irq_period);
+       period = max_t(u64, 10000, counter->hw.sample_period);
        hrtimer_forward_now(hrtimer, ns_to_ktime(period));
 
        return ret;
@@ -2959,7 +2959,7 @@ static void perf_swcounter_add(struct perf_counter *counter, u64 nr,
 {
        int neg = atomic64_add_negative(nr, &counter->hw.count);
 
-       if (counter->hw.irq_period && !neg && regs)
+       if (counter->hw.sample_period && !neg && regs)
                perf_swcounter_overflow(counter, nmi, regs, addr);
 }
 
@@ -3080,8 +3080,8 @@ static int cpu_clock_perf_counter_enable(struct perf_counter *counter)
        atomic64_set(&hwc->prev_count, cpu_clock(cpu));
        hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
        hwc->hrtimer.function = perf_swcounter_hrtimer;
-       if (hwc->irq_period) {
-               u64 period = max_t(u64, 10000, hwc->irq_period);
+       if (hwc->sample_period) {
+               u64 period = max_t(u64, 10000, hwc->sample_period);
                __hrtimer_start_range_ns(&hwc->hrtimer,
                                ns_to_ktime(period), 0,
                                HRTIMER_MODE_REL, 0);
@@ -3092,7 +3092,7 @@ static int cpu_clock_perf_counter_enable(struct perf_counter *counter)
 
 static void cpu_clock_perf_counter_disable(struct perf_counter *counter)
 {
-       if (counter->hw.irq_period)
+       if (counter->hw.sample_period)
                hrtimer_cancel(&counter->hw.hrtimer);
        cpu_clock_perf_counter_update(counter);
 }
@@ -3132,8 +3132,8 @@ static int task_clock_perf_counter_enable(struct perf_counter *counter)
        atomic64_set(&hwc->prev_count, now);
        hrtimer_init(&hwc->hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
        hwc->hrtimer.function = perf_swcounter_hrtimer;
-       if (hwc->irq_period) {
-               u64 period = max_t(u64, 10000, hwc->irq_period);
+       if (hwc->sample_period) {
+               u64 period = max_t(u64, 10000, hwc->sample_period);
                __hrtimer_start_range_ns(&hwc->hrtimer,
                                ns_to_ktime(period), 0,
                                HRTIMER_MODE_REL, 0);
@@ -3144,7 +3144,7 @@ static int task_clock_perf_counter_enable(struct perf_counter *counter)
 
 static void task_clock_perf_counter_disable(struct perf_counter *counter)
 {
-       if (counter->hw.irq_period)
+       if (counter->hw.sample_period)
                hrtimer_cancel(&counter->hw.hrtimer);
        task_clock_perf_counter_update(counter, counter->ctx->time);
 
@@ -3223,7 +3223,7 @@ static const struct pmu *tp_perf_counter_init(struct perf_counter *counter)
                return NULL;
 
        counter->destroy = tp_perf_counter_destroy;
-       counter->hw.irq_period = counter->hw_event.irq_period;
+       counter->hw.sample_period = counter->hw_event.sample_period;
 
        return &perf_ops_generic;
 }
@@ -3323,15 +3323,15 @@ perf_counter_alloc(struct perf_counter_hw_event *hw_event,
        pmu = NULL;
 
        hwc = &counter->hw;
-       if (hw_event->freq && hw_event->irq_freq)
-               hwc->irq_period = div64_u64(TICK_NSEC, hw_event->irq_freq);
+       if (hw_event->freq && hw_event->sample_freq)
+               hwc->sample_period = div64_u64(TICK_NSEC, hw_event->sample_freq);
        else
-               hwc->irq_period = hw_event->irq_period;
+               hwc->sample_period = hw_event->sample_period;
 
        /*
-        * we currently do not support PERF_RECORD_GROUP on inherited counters
+        * we currently do not support PERF_SAMPLE_GROUP on inherited counters
         */
-       if (hw_event->inherit && (hw_event->record_type & PERF_RECORD_GROUP))
+       if (hw_event->inherit && (hw_event->sample_type & PERF_SAMPLE_GROUP))
                goto done;
 
        if (perf_event_raw(hw_event)) {